基于超越阈值模型的低矮房屋实测风压极值估计方法

台风几乎每年都造成我国低矮房屋的巨大破坏和人员伤亡，了解屋面风压极值，尤其是屋面的角部、边缘及屋脊位置的风压极值分布具有重要意义。本文采用极值理论下的超越阈值模型来合理估计低矮房屋在台风影响下屋面实测风压极值。首先推导了描述经验平均超额分布与阈值关系的理论公式，同时以广义Pareto分布作为风压极值的拟合分布。然后，依托同济大学浦东风荷载实测基地，以低矮房屋在2016年台风”马勒卡”作用下屋面风压实测数据为样本，比较不同阈值对估计结果的影响。最后，以标准时距下多次观察极值的平均值作为标准极值，分析了本文方法和常用方法的极值估计结果与标准极值之间的误差，其结果表明，本文方法的误差在5%以内，尤其对屋面极值风压较大的区域估计结果较为理想。
Abstract：Typhoons have caused great damage to low-rise buildings and significant casualties in China almost every year.To understand the probability distribution of extreme values of wind pressures on roofs,especially on the corner,edge and ridge positions,is of important significance.The Peaks over Threshold (POT) model in the classical extreme value theory was used to properly estimate the extreme values of wind pressures on roofs of low-rise buildings under typhoons.First, theoretical formulas were derived,which describe the relation between the mean excess distribution and the threshold.Meanwhile,the generalized Pareto distribution was used as the fitted distribution of the wind pressure extreme values.Next,based on the measured roof pressures data on low-rise buildings collected under the Typhoon Malakas in the Pudong measurement base of Tongji University,the estimation results using different thresholds were compared to study the influence of the threshold.Then,the mean value of measured extreme values repeatedly observed in a standard time interval was defined as a standard extreme value to compare the estimated results by the POT model method and by other traditional methods,and the errors between them were analyzed.The research shows that the errors of the POT model method are less than 5%,and the POT model method is better than other methods,particularly when estimating larger extreme wind pressures on roofs. 